Professional Documents
Culture Documents
New Metallic Damper With Multiphase Behavior For Seismic Protection of Structures
New Metallic Damper With Multiphase Behavior For Seismic Protection of Structures
Abstract: This paper proposes a new metallic damper based on the plastic deformation of mild steel.
It is intended to function as an energy dissipation device in structures subjected to severe or extreme
earthquakes. The damper possesses a gap mechanism that prevents high-cycle fatigue damage un-
der wind loads. Furthermore, subjected to large deformations, the damper presents a reserve of
strength and energy dissipation capacity that can be mobilized in the event of extreme ground mo-
tions. An extensive experimental investigation was conducted, including static cyclic tests of the
damper isolated from the structure, and dynamic shake-table tests of the dampers installed in a
reinforced concrete structure. Four phases are distinguished in the response. Based on the results of
the tests, a hysteretic model for predicting the force-displacement curve of the damper under arbi-
trary cyclic loadings is presented. The model accurately captures the increment of stiffness and
strength under very large deformations. The ultimate energy dissipation capacity of the damper is
found to differ depending on the phase in which it fails, and new equations are proposed for its
prediction. It is concluded that the damper has a stable hysteretic response, and that the cyclic be-
havior, the ultimate energy dissipation capacity and failure are highly predictable with a relatively
Citation: Benavent-Climent, A.; simple numerical model.
Escolano-Margarit, D.;
Arcos-Espada, J.; Ponce-Parra, H. Keywords: metallic damper; mild steel; shake-table test; cyclic loading; energy dissipation
New Metallic Damper with
Multiphase Behavior for Seismic
Protection of Structures. Metals 2021,
11, 183. https://doi.org/
1. Introduction
10.3390/met11020183
The 1994 Northridge (California) and 1995 Kobe (Japan) earthquakes highlighted
Academic Editor: Ravi Kiran that a conventional seismic design—where the beams and columns of the main structure
Yellavajjala are designed to dissipate energy through plastic deformations under a severe ground mo-
Received: 1 January 2021 tion—results in significant structural and nonstructural damage and the interruption of a
Accepted: 19 January 2021 building’s use after the event. Since the beginning of the 21st century, seismic engineering
Published: 20 January 2021 has undergone a transition toward so-called Performance-Based Design (PBD), aimed at
controlling/minimizing the consequent damage and financial losses. Structures with en-
Publisher’s Note: MDPI stays neu-
ergy dissipation systems have proven to be a very effective solution to attain the objectives
tral with regard to jurisdictional
of PBD. They consist of a main structure that supports the gravity loads and an energy
claims in published maps and insti-
dissipation system working in parallel. The latter is formed by special structural elements
tutional affiliations.
called energy dissipation devices (EDDs), or simply dampers, plus the auxiliary elements
that connect the EDDs with the main structure. The EDDs are in charge of absorbing most
of the energy input by the earthquake, releasing the main structural elements from dissi-
Copyright: © 2021 by the authors.
pating energy through plastic deformations. During the earthquake, the response of the
Licensee MDPI, Basel, Switzerland. main structure is essentially elastic, and at the end of the event it is basically undamaged.
This article is an open access article The damage concentrates in the EDDS, which is purposely designed to be easily in-
distributed under the terms and con- spected, replaced or repaired after a severe (commonly called “design earthquake”) or an
ditions of the Creative Commons At- extreme (“maximum credible earthquake”) ground motion. This allows for the continu-
tribution (CC BY) license (http://cre- ous use of a building without interruption, enhancing resilience. Since its first application
ativecommons.org/licenses/by/4.0/).
in the early 1970s [1], the addition of EDDs has proven to be an effective technology for
the seismic protection of buildings. EDDs can be classified as either displacement-depend-
ent or velocity-dependent. The former includes metallic dampers (also called hysteretic
dampers) and friction dampers [2–5]. This paper is focused on metallic EDDs, whose
source of energy dissipation is the yielding of metals [6]. They are built with well-known
and reliable material (mild steel) that have a stable hysteretic behavior and a large inher-
ent plastic deformation capacity. A comprehensive state-of-art review of the development
and implementation of metallic EDDs can be found in [7]. Different types of metallic EDDs
have been proposed in the literature and used in practical applications. Among the most
popular are the Added Damping and Stiffness damper (ADAS) [8] and its triangular ver-
sion (TADAS) [9], the buckling restrained brace [10], or the steel plate with slits [11], also
called slit-type damper herein. The latter consists of strips of steel having constant or var-
iable width, made by opening slits—either simple (Figure 1a) or double (Figure 1b)—in a
steel plate. The mechanism for energy dissipation resides in plastic bending/shearing de-
formations of the steel strips.
The slit-type damper has been successfully implemented in actual building designs
such as the Sapporo Hotel in Japan [12] and several studies have been carried out over the
last decade. Chan and Albermani [6] proposed a slit-type damper that is fabricated from
a standard structural wide-flange section by opening slits in the web. Oh et al. [13] verified
through cyclic tests the seismic performance of steel structures with slit-type dampers at
the bottom flange and confirmed their satisfactory performance. Striving to optimize the
EDD, Ghabraie et al. [14] and Zheng et al. [15] investigated new shapes for the steel strips
(i.e., with a variable section) to enhance their energy dissipation capacity. Lee et al. [16]
tested the cyclic performance of three different shapes in order to reduce stress concentra-
tion: dumbbell-shaped strip, tapered strip and hourglass-shaped strip. Amiri et al. [17]
studied a block slit damper with a very low height-to-thickness ratio. Shao et al. [18] in-
vestigated the double slit configurations (spine) shown in Figure 1b. Benavent–Climent
[19] proposed a tube-in-tube damper based on flexural/shear yielding of the steel strips
formed by opening slits on the walls of hollow structural sections (Figure 1c). A similar
concept was applied by Lee and Kim [20], who proposed a box-shaped steel slit damper.
In the last ten years, hybrid dampers that use two different passive elements combined in
a single device have been proposed, e.g., viscoelastic dampers and metallic dampers [21],
or friction dampers and metallic dampers [22]. Nowadays, a limited body of work exists
for hybrid dampers, but it is an interesting solution that receives increasing attention [23].
Figure 1. Deformed shapes of slit-type steel plates with (a) simple or (b) double configuration, and (c) steel tubes with
openings in the walls.
negligible, so that it can be endured by the main structure with no need for dampers [26].
This paper presents a new damper that avoids the problem of high cycle fatigue by intro-
ducing a gap mechanism. The proposed damper is intended to be used in low-to-mid-rise
buildings (up to about 12 floors) designed following modern codes, that can endure wind
loads with no need for braces.
Another characteristic feature of most metallic dampers is that beyond a given dis-
placement (typically large), they present a significant increase of the restoring force and
plastic stiffness. In the case of the slit-type damper, this occurs when the displacement of
the ends of the steel strip in the direction of its axis are restrained. This restriction gives
rise to important axial forces in the steel strips (in addition to bending and shear forces),
that result in the aforementioned significant increase of the restoring force and stiffness.
This is not exclusive of slit-type dampers; it occurs with other types of metallic dampers
(e.g., the TADAS damper [9]) and can be prevented using appropriate connections (e.g.,
using pin-connections and slotted holes) at the expense of incrementing the production
cost of the damper. The increment of restoring forces at large displacements can be con-
sidered as a flaw from the standpoint of the additional forces that this overstrength im-
poses upon the main structure and foundation. Yet if it is anticipated on design, and the
main structure is prepared for it, it may prove beneficial as a reserve of strength and en-
ergy dissipation capacity, being necessary in the case of an extremely high amplitude
earthquake. In fact, as shown later in this paper, the reserve of energy dissipation capacity
of the damper is very large when the restoring force starts to increase significantly until
failure. An additional reason why this increase in the restoring force under large displace-
ments is often ignored is that it cannot be easily captured by the numerical models typi-
cally used to characterize the hysteretic behaviour of metallic dampers (i.e., the Bouc–Wen
model). This paper presents a simple numerical model that can accurately reproduce the
increment of restoring force at large displacements, as well as the amount of dissipated
energy, and predict the force displacement hysteretic curves of the metallic damper under
arbitrarily applied cyclic loading until failure.
from one phase to the following can be determined upon design by using appropriate
values for 𝛿 , for the geometry and number of steel strips and for the geometry of the
outer tube. The values of the displacements can be tuned so that under wind loads the
damper’s response is within Phase I, under moderate (frequent) earthquakes it does not
exceed Phase II, under a severe (“design earthquake”) motion the response does not go
beyond Phase III, and under a rare (“maximum credible earthquake”) ground motion the
damper enters Phase IV.
Figure 2. Concept of the damper: (a) section A–A’; (b) elevation; (c) plan.
2.2. Materialization
The design concept explained above is materialized in the metallic damper shown in
Figure 3a. It is made of two standard hollow structural sections. In two faces of the outer
tube (Figure 3b), slotted holes are opened using a waterjet cutting system in order to have
smooth finished surfaces and to prevent altering the properties of the steel by heat. The
steel strips between the slotted holes have a spinal configuration. To prevent out-of-plane
buckling of the strips, the central part of the spine is strengthened with a rectangular plate
fixed with pre-stressed high strength bolts to the outer tube. The inner tube (Figure 3c)
has four stoppers fixed with high strength steel rods that are post-tensioned in order to
avoid any slippage of the stopper with respect to the inner tube. The well-pondered loca-
tion of these stoppers allows for the gap 𝛿 . Figure 3d offers a photo of the metallic damp-
ers used in the test campaign explained next. Figure 4 reflects the implementation of the
metallic damper in a frame structure.
Figure 3. Materialization of the metallic damper: (a) assembled damper; (b) outer tube; (c) inner
tube; (d) photograph of the specimens used in the tests.
3. Experimental Research
The performance of seven identical MP-TTD specimens, referred to as MP-TTD -0 to
MP-TTD-6 herein, up to failure was studied experimentally under quasi-static cyclic tests,
dynamic seismic shake table tests, and a mixture of dynamic plus quasi-static tests. For all
test types, failure was assumed to occur when the restoring force opposed by the damper
started to decrease under increasing imposed deformations.
Specimen MP-TTD-0 was tested isolated from the structure under quasi-static cyclic
loadings. Specimens MP-TTD-1 to MP-TTD-6 were installed in a reinforced concrete (RC)
structure that was subjected to realistic seismic loadings on a shake table. During the dy-
namic shake table tests, specimens MP-TTD-2 and MP-TTD-3 reached failure, specimens
MP-TTD-1 and MP-TTD-4 suffered severe plastic deformations, and specimens MP-TTD-
5 and MP-TTD-6 remained within the elastic range. After the dynamic shake table tests,
specimens MP-TTD-1, MP-TTD-4, MP-TTD-5 and MP-TTD-6 were removed and isolated
from the RC structure, then subjected to additional quasi static cyclic loading tests until
failure.
3.1. Description of MP-TTDs Tested, Material Properties and Predicted Axial Strength and
Yield Deformation
All dampers were made from the same hollow tubes in order to ensure identical char-
acteristics for the steel. The nominal dimensions of the outer tube were #200.120.4 (width,
depth, and thickness in mm), while the inner tube was #180.100.4. Figure 5 shows the
geometry of the tested specimens, including detailed geometry of the steel strips. The
specimens represent at a 2/5 scale the dampers that would be installed in a full-scale struc-
ture having 6500 mm span length and 3800 mm story height. As shown in Figure 5, the
value selected for the gap was 𝛿 = 5 mm. The axial deformation 𝛿 = 5 mm in the
scaled damper corresponds to an inter-story drift of 0.38%. This inter-story drift is smaller
than the maximum value (about 0.5%) that typical reinforced concrete or steel structures
can endure in the elastic range. Therefore, the scaled damper was designed with a gap of
𝛿 = 5 mm because typical structures can endure the lateral displacements associated
with this gap without damage. In practical application, for fastening the assembly of the
inner and outer tubes so that the two gaps are equal, the brace damper would be supplied
with provisional and easily removable steel blocks that close the gap. These steel blocks
would be removed once both ends of the brace damper are fixed to the main structure.
Metals 2021, 11, 183 6 of 30
The steel class was S-275JR. The material properties were determined from three cou-
pon tensile tests. Table 1 summarizes the mean Young modulus E, yield stress σy, ultimate
stress σu, and the corresponding strains εy and εu.
E σy σu εy εu
(GPa) (MPa) (MPa) (%) (%)
210 362 530 0.349 9.702
Assuming that the ends of the steel strips are perfectly clamped (no rotation), and
replacing the total height of the strip h + 2r (see Figure 5d) by an equivalent height h′ given
by h′ = h + (2r2/(h + 2r)) to take into account the rounded ends [19]. The meanings of h (=65
mm) and r (=5 mm), are shown in the detail of the strip of Figure 5d. The mechanical
properties of the MP-TTD can be predicted through simple mechanical principles. Con-
sidering an MP-TTD constituted of n steel strips, the force Q (see Figure 1b) when all fibers
of the cross-section of the strips reach σy (referred to as yield strength Qy hereafter), and
the force when all fibers of the cross-section reach σu, (referred to as strength QB herein)
are given by [19]:
𝜎 𝑡𝑏 𝜎 𝑡𝑏
𝑄 =𝑛 ; 𝑄 =𝑛 (1)
2ℎ′ 2ℎ′
The meanings of t (= 4 mm) and b (= 5 mm) are shown in the detail of the strip of
Figure 5d. The yield displacement 𝛿 of a MP-TTD with n steel strips can be estimated
as follows [19]:
𝑄 ℎ′ ℎ + 2𝑟
𝛿 = 1 + 3 ln (2)
𝑛𝐸𝑡𝑏 ℎ′
Using the material properties of Table 1 and the above equations, the predicted
strengths and axial displacement of the specimens are Qy = 23.15 kN, QB = 33.89 kN and
𝛿 = 1.06 mm.
Metals 2021, 11, 183 7 of 30
Figure 5. MP-TTD damper specimen: (a) section A-A’; (b) elevation; (c) detail of the stoppers; (d) detail of a single steel
strip; (e) cross-section; (f) cross-section; (g) detail of the part of the outer tube that restrains the movement of the ends of
the strips along their axes. Units in mm.
TTD-5 and 𝜙 = 1 for MP-TTD-6), and the number of repetitions per amplitude applied
(two for specimens MP-TTD-0 and MP-TTD-6, one for specimen MP-TTD-5).
Actuator
Load cell
Outer tube
LVDT
Inner tube
(a) (b)
Figure 6. Experimental set-up for quasi-static tests: (a) components, (b) photograph.
constant the restoring force Q until point 5. Sixth, when the deformations are very large,
the restoring force Q starts to increase significantly (segment 5–6). The latter is caused by
restriction on the movements of the ends of the strips in the direction perpendicular to the
applied force Q (i.e., in the direction of the strip’s axis). In the specimens tested, the last
(sixth) effect appears for axial deformations beyond approximately 20 mm, which corre-
sponds to an inter-story drift of about 1.5% in a conventional frame structure. The defor-
mation associated with the onset of the overstrength (i.e., point 5 in Figure 11) depends
on the level of restriction of the relative movements of the ends of the strips perpendicular
to the axis of the MT-TTD, which can be tuned by controlling the stiffness of the lateral
parts of the outer tube; that is, the inertia of the elements shown in Figure 5g with respect
to the x–x′ axis passing through its centroid. Figures 8–10 show with dot lines the force-
displacement relationships after the peak strength is reached. It is worth noting that the
degradation of strength after the assumed point of failure is gradual; this is due to the
successive (i.e., not simultaneous) failure of the steel strips.
508
A
B
278
1510
MP-TTD-06
6
MP
-TT
D-0
4
3016
2000
A'
-05
-TTD
MP
6
1510
278
(a)
(b)
Figure 12. Shake table test experimental set-up: (a) aerial plan view; (b) elevation. Units in mm.
Metals 2021, 11, 183 12 of 30
Figure 13. Photograph of the RC structure with the MP-TTDs before the tests.
(a) (b)
Figure 14. History of accelerations applied to the shake table in X (a) and Y (b) directions.
Since the RC frame was not severely damaged and retained its capacity to sustain the
vertical gravity loads, and since none of the dampers reached failure, it was decided that
the seismic shakings should continue with the additional test, C400. It is worth recalling
that C400 corresponds to a very rare earthquake, with higher intensity than the “maxi-
mum considered earthquake” prescribed by seismic codes for structures of normal im-
portance (e.g., residential buildings). The return period of the earthquake simulated by
test C400 (RP = 2687 years) is close to the value assigned by the European seismic code
Eurocode 8 (2500 years) for verifying the “near collapse” limit of buildings pertaining to
consequence class CC3a (i.e., buildings whose seismic resistance is of great importance in
view of the consequences, e.g., schools, assembly halls, cultural institutions, etc.) During
test C400, dampers MP-TTD-2 (Figure 18c) and MP-TTD-3 (Figure 19c) reached their ulti-
mate energy dissipation capacity and failed. Both entered clearly in Phase IV of their
multi-phase behaviour, as evidenced by the significant increases of strength and plastic
stiffness. Before failure, dampers MP-TTD-2 and MP-TTD-3 were able to sustain ex-
tremely large axial deformations (39 mm MP-TTP-2 and 30 mm MP-TTD-3), up to 37 and
28 times the yield deformation, and develop very large forces of 145 kN (MP-TTD-2) and
110 kN (MP-TTD-3), which are about six times the nominal yield force Qy (=23.15 kN).
These very large forces are attributed to the development of large axial forces along the
axis of the steel strips when the damper is subjected to large deformations along its axis.
The maximum deformation 𝛿 = 39 mm corresponds to an inter-story drift of 3%
(=100(𝛿 /𝐻 𝑐𝑜𝑠𝛼))—that is, above the value (2.5%) specified by SEAOC [29] as indica-
tive of collapse. Inspection of the RC frame after test C400 revealed extensive plastic de-
formations at column ends, but the RC structure did not collapse. In sum, the overstrength
and the associated increase of the plastic stiffness, together with the additional energy
dissipation capacity provided by the dampers in Phase IV, prevented the RC structure
from collapse under a ground motion of extreme severity with a PGA as large as 0.62 g.
Dampers MP-TTD-1 (Figure 16c) and MP-TTD-4 (Figure 17c) also experienced large plas-
tic deformations and maximum axial displacements up to 𝛿 = 17 mm, but they did
not fail; they remained in Phase III. For this reason, in order to investigate their ultimate
energy dissipation capacity, it was decided to subject dampers MP-TTD-1 and MP-TTD-4
to additional quasi-static cyclic deformations until failure, as explained below.
(a)
Metals 2021, 11, 183 15 of 30
(b)
(c)
(d)
Figure 16. Response of damper MP-TTD-1 during tests: (a) C200, (b) C300, (c) C400 and (d) quasi-static test.
(a)
(b)
Metals 2021, 11, 183 16 of 30
(c)
(d)
Figure 17. Response of damper MP-TTD-4 during tests: (a) C200, (b) C300, (c) C400, and (d) quasi-
static.
(a)
(b)
Metals 2021, 11, 183 17 of 30
(c)
Figure 18. Response of damper MP-TTD-2 during tests: (a) C200, (b) C300 and (c) C400.
(a)
(b)
(c)
Figure 19. Response of damper MP-TTD-3 during tests: (a) C200, (b) C300 and (c) C400.
Metals 2021, 11, 183 18 of 30
(a) (b)
Figure 20. Additional quasi-static loading histories: (a) MP-TTD-1 and (b) MP-TTD-4.
4. Discussion
4.1. Decomposition into Skeleton and Bauschinger Parts
Previous research on metallic structural elements [30] has shown that the ultimate
energy dissipation capacity is path-dependent, that is, it varies with the loading pattern
applied. In the case of seismic loadings, the cyclic loading pattern that a future earthquake
will impose upon a metallic damper cannot be foreseen, because the history of ground
acceleration in itself is unpredictable. A convenient way to address this load-path depend-
ency and characterize the energy dissipation capacity of the damper consists of decom-
posing the hysteretic loops into two parts: the so-called skeleton part and the Bauschinger
part [31]. Let us consider the typical hysteresis Q-δ curve obtained for a damper until fail-
ure that is shown in Figure 21a. To simplify the explanation, and given that the presence
of a gap, 𝛿 does not affect the decomposition, in the following explanation it is consid-
ered that 𝛿 = 0.
Within each domain of loading (positive or negative), the skeleton is defined through
a sequential connection of the segments that exceed the maximum load Q attained by the
metallic element in previous cycles of the same load domain. As depicted in Figure 21a,
that would be the blue fragments 0–1, 5–6, 11–12, and 17–18 for positive loading, and 2–3,
8–9, and 14–15 for negative loading. Reassembling all the extracted skeleton parts succes-
sively in a continuous plot, the skeleton curve shown in Figure 21b is retrieved. Kato and
Akiyama [31] showed that the shape of the skeleton curve is independent of the charac-
teristics of the history of loading applied; but the end points, that is, the maximum accu-
mulated deformations 𝛿 , 𝛿 , do indeed depend on the loading history. Furthermore,
these authors demonstrated experimentally that the shape of the skeleton part approxi-
mately coincides with the Q-δ relationship that would be obtained under monotonic load-
ing. Given the independency of the shape of the skeleton part from the history of loading,
the shape of the skeleton curves obtained for each damper were approximated in this
study by a single pentalinear curve that is shown with dash and dot lines in Figure 21b.
This pentalinear curve is characterized by Qy = 23.15 kN, QB = 33.89 kN, δy = 1.06 mm, Ke =
Qy/δy = 21.83 kN/mm, Kp1 = 8.73 kN/mm; and Kp2 = 1.56 kN/mm, Kp3 = 3.64 kN/mm, Kp4 =
Metals 2021, 11, 183 19 of 30
14.55, QB2 = 50.1 kN, QB3 = 96.4 kN. Referring to Figure 21b, the deformations 𝛿 , 𝛿 ,
𝛿 corresponding to the end of the segments with stiffness Kp1, Kp2, Kp3, respectively, are
𝛿 = 2.3 𝑚𝑚, 𝛿 = 12.9 𝑚𝑚 and 𝛿 = 25.6 𝑚𝑚. The values of Qy, QB and 𝛿 ,
can be predicted with Equations (1) and (2). For convenience, the approximated pentalin-
ear skeleton curve can be expressed in nondimensional form in terms of the following
parameters:
𝑘 = ; 𝑘 = ; 𝑘 = ; 𝑘 = ; 𝜏 = ; 𝜏 = ; 𝜏 = (3)
For the dampers tested in this study, their values are: kp1 = 2/5, kp2 = 1/14, kp3 = 1/6, kp4
= 2/3, 𝜏 = 1.46, 𝜏 = 𝜏 + 10𝑘 , 𝜏 = 𝜏 + 10𝑘 + 12𝑘 .
The areas enveloped by the skeleton curves in each domain of loading (i.e., the
shaded areas in Figure 21b) until 𝛿 , 𝛿 will be denoted by SWu+ and SWu− and repre-
sent the energy dissipated by the damper on the skeleton part, in each domain of loading.
Since 𝛿 , 𝛿 depend on the history of loading, SWu+ and SWu− are different for each
damper tested.
The Bauschinger part comprises the remaining segments of the Q-δ curve which,
starting at Q = 0, seek the point corresponding to the maximum load attained in the pre-
ceding cycles for each loading domain. In Figure 21a that would be the green fragments
4–5, 10–11, and 16–17 for positive loading, and 7–8, and 13–14 for negative loading. Figure
21c shows all the Bauschinger parts extracted from Figure 21a and plotted correlatively.
The sum of the areas enveloped by the Bauschinger parts in each domain of loading, BWu+
and BWu−, represents the fraction of the total plastic strain energy dissipated by the damper
that is consumed on the Bauschinger part. The unloading branches in both the skeleton and
Bauschinguer parts (i.e., segments 1–2, 6–7, 12–13, 18–19, 3–4, 9–10, and 15–16 in Figure 21b,c)
have the initial elastic stiffness Ke.
Metals 2021, 11, 183 20 of 30
Figure 21. Decomposition of Q-δ: (a) overall curve, (b) skeleton part and (c) Bauschinger part.
The total energy dissipation in its nondimensional form for skeleton, Bauschinger
and overall can be simply assessed as follows:
𝜂= 𝜂 + 𝜂 ; 𝜂= 𝜂 + 𝜂 ; 𝜂= 𝜂 + 𝜂 ; 𝜂= 𝜂+ 𝜂 (5)
Following the procedure explained in Section 4.1, the Q-δ hysteresis loops obtained
for each damper until failure were decomposed into the skeleton and Bauschinger parts.
The values for the nondimensional ratios defined in Equations (3)–(5) were obtained and
are summarized in Table 2 for the dampers that did not enter Phase IV, and in Table 3 for
the dampers that entered Phase IV. Figure 22 plots the discrete values of 𝜂, 𝜂, and 𝜂
against 𝜂. Both the dampers that did not enter Phase IV and those that entered Phase
IV are represented together for comparison purposes.
Figure 22a shows the discrete values of 𝜂 against 𝜂 obtained from the tests.
Since the relation between 𝜂 against 𝜂 depends only on the shape of the skeleton
curve, it can be easily predicted from the pentalineal approximation of the skeleton curve
adopted in Section 4.1 (see Appendix). The 𝜂 − 𝜂 curve predicted in this way is plot-
ted in Figure 22a, with a solid bold line, and is given by:
For 𝜂 ≤ 𝜂 :
𝑘
𝜂=2 𝜂+ 𝜂 (6)
1−𝑘
For 𝜂 ≤ 𝜂 ≤ 𝜂 :
𝑘 ( 𝜂− 𝜂 )
𝜂= 𝜂 (1 + 𝜏 ) + ( 𝜂 − 𝜂 ) 2𝜏 + (7)
(1 − 𝑘 )
For 𝜂 ≤ 𝜂 ≤ 𝜂 :
𝑘 𝜂 − 𝜂
𝜂= 𝜂 (1 + 𝜏 ) + 𝜂 − 𝜂 2𝜏 +
1−𝑘
(8)
𝑘 𝜂− 𝜂
+ 𝜂− 𝜂 2𝜏 +
1−𝑘
For 𝜂 > 𝜂 :
𝑘 𝜂 − 𝜂
𝜂= 𝜂 (1 + 𝜏 ) + 𝜂 − 𝜂 2𝜏 + +
1−𝑘
(9)
𝑘 ( 𝜂 − 𝜂 ) 𝑘 ( 𝜂 − 𝜂 )
( 𝜂 − 𝜂 ) 2𝜏 + +( 𝜂 − 𝜂 ) 2𝜏 +
(1 − 𝑘 ) (1 − 𝑘 )
Note that both the dampers not entering Phase IV and those entering Phase IV lie
approximately on the same 𝜂 − 𝜂 curve. This is because the shape of the skeleton
curves is similar for all dampers, and 𝜂 is uniquely determined by the amount of defor-
mation accumulated on the skeleton part 𝜂.
Likewise, Figure 22b plots the discrete values of 𝜂 against 𝜂 . Two groups of
points must be distinguished here. The first are the points corresponding to the dampers
that did not enter Phase IV, and the second group those of the dampers that entered Phase
IV. The points of the first group lie approximately on a line defined by:
𝜂=𝑎∙ 𝜂+𝑏 (10)
With a = −18.4 and b’ = 965. On the other hand, the points of the second group lie on
a line parallel to that of the first group of points, but displaced upward, vertically; the
equation of the second line being determined by Equation (10) using a = −18.4 and b’ =
1295. This means that, for similar deformations accumulated on the skeleton part, the
Metals 2021, 11, 183 22 of 30
dampers that enter Phase IV dissipated about 35% more energy in the Bauschinger part
that those that did not reach Phase IV.
Finally, Figure 22c shows the total ultimate energy dissipation capacity. Since the
curve corresponding to the Bauschinger part is different for the dampers that entered
Phase IV and for those that did not, two curves (in red) are obtained for the normalized
ultimate energy dissipation capacity. These curves can be used to predict the failure of the
damper when subjected to arbitrarily applied loading (as would be induced by earth-
quakes), to be explained in the model proposed in the next section.
(a)
(b)
(c)
Figure 22. Ultimate energy dissipation capacity: (a) skeleton part, (b) Bauschinger part, and (c)
total energy dissipated.
Table 2. Ultimate energy dissipation capacity of the dampers that did not enter Phase IV.
Specimen Loading δ
S u+ δ
S u− η
ep + η
ep − η
S + η
S − η
B + η
B − η S Bη η
MP-TTD 0 Quasi-Static 12.69 13.08 10.94 11.30 21.96 22.99 213.15 235.47 44.95 448.62 493.57
MP-TTD 5 Quasi-Static 21.80 18.33 19.50 16.24 41.72 36.75 113.20 142.09 78.47 255.29 333.76
MP-TTD 6 Quasi-Static 15.06 9.45 13.16 7.89 24.27 17.05 293.86 309.69 41.32 603.55 644.86
MP-TTD 1 Mixed 6.24 10.43 4.87 8.81 10.78 19.09 376.73 389.42 29.87 766.15 796.02
MP-TTD 4 Mixed 22.13 20.60 19.81 18.38 46.49 51.78 142.47 195.73 98.27 338.21 436.47
Table 3. Ultimate energy dissipation capacity of the dampers that entered Phase IV.
Specimen Loading δ
S u+ δ
S u− η
ep + η
ep - η
S + η
S − η
B + η
B − Sη B η η
MP-TTD 2 Dynamic 7.06 28.96 5.64 26.23 8,90 87.38 386.41 327.91 96.28 714.32 810.61
Metals 2021, 11, 183 23 of 30
MP-TTD 3 Dynamic 11.58 18.82 9.89 16.7 18.72 48.56 483.58 317.26 67.28 800.84 868.13
5. Model for Predicting the Hysteretic Curve under Arbitrary Cyclic Loading and Fail-
ure
The same rationale as explained in Section 4.1—to deconstruct the force displacement
curve Q-δ obtained by subjecting a metallic damper to arbitrary cyclic loadings until fail-
ure—can be applied to construct (i.e., to predict) the hysteretic Q-δ loops developed by
the damper if it is subjected to arbitrary cyclic loadings. It can be attained by means of a
simple polygonal hysteretic model that requires characterization of the shape of the skel-
eton and Bauschinger parts. This is accomplished in the next subsections. Aside from its
simplicity (easy implementation for example in subroutines for conducting non-linear
time history analyses), the main advantages of the polygonal hysteretic model proposed
next with respect to other well-known smooth hysteretic models based on the Bouc–Wen
formulations are: (i) it can accurately reproduce the damper’s energy-consumption path
along the skeleton and Bauschinger parts, and is therefore able to predict the failure of the
damper; and (ii) it can accurately capture the significant increment of strength and stiff-
ness in the range of large deformations.
𝛿=𝛽 𝛿 (11)
With 𝛽 = 0.9. Secondly, for a given Bauschinger segment whose maximum force at
the end point is Qmax, the value of 𝑄 was computed, making the area below the actual
Bauschingher segment (bold line in Figure 23) and the area below the bilinear approxima-
tion (dash line in Figure 23) equal. The pair of values (𝑄 , 𝑄 ) calculated in this way for
the Bauschinger segments obtained from the tests are plotted in Figure 24b. They lie ap-
proximately in a line given by the following expression, where 𝛼 = 0.5:
𝑄 = 𝛼𝑄 (12)
Metals 2021, 11, 183 24 of 30
(a) (b)
Figure 24. Parameters that characterize the shape of Bauschinger segments: (a) 𝛽; (b) 𝛼.
prescribed by Equation (12). From point 11 onward the damper starts consuming the skel-
eton part again, and the Q-δ curve coincides with segment 11–12 of Figure 25a.
Figure 25. Construction of the hysteretic curve: (a) skeleton curve, (b) load history and (c) hysteretic curve.
The energy consumption path followed by the damper under the loading history of
Figure 25b can be plotted in the 𝜂 − 𝜂 plane of Figure 21c, which is redrawn in Figure
26 for convenience. The energy dissipated by the damper from point 1 to 4 in Figure 25c
is the area below the Q-δ curve, W4, that can be expressed in nondimensional form by
𝜂 = 𝑊 /(𝑄 𝛿 ). Meanwhile, the deformation accumlated on the skeleton part up to point
4 in a nondimensional form is 𝜂 = 𝛿 /𝛿 . The point of coordinates ( 𝜂 , 𝜂 ) is
shown in Figure 26. Since from point 1 to 4 all the energy has been consumed entirely by
the skeleton part, the segment representing the load path 1–4 in the ( 𝜂 , 𝜂 ) space
follows the curve that represents the energy dissipated by the skeleton part. From point 4
to 6, the damper dissipates an increment of energy 𝛥𝜂 = 𝑊 /(𝑄 𝛿 ), where 𝑊 is
the area of the Q-δ curve between points 4 and 6 in Figure 25c. Since between these points
all the dissipated energy consumed only the Bauschinger part, then 𝛥 𝜂 = 0 and the
segment 4–6 in the ( 𝜂 , 𝜂 ) space is a vertical line. The end point of coordinates ( 𝜂 ,
𝜂 ) is shown in Figure 26. From point 6 to 8, the damper dissipates an increment of en-
ergy 𝛥𝜂 = 𝑊 /(𝑄 𝛿 ), where 𝑊 is the area of the Q-δ curve between points 6 and
8 in Figure 25c. Between these points, all the dissipated energy consumed only the skele-
ton part with 𝛥 𝜂 = 𝛿 /𝛿 , then the segment 6–8 in the ( 𝜂 , 𝜂 ) space is par-
alell to the curve that represents the energy dissipated by the skeleton part. The end point
of coordinates ( 𝜂 , 𝜂 ) is shown in Figure 26. Following this procedure, the energy
consumption path in the ( 𝜂 , 𝜂 ) space can be traced, and the damper will fail when
the curve that represents the ultimate energy dissipation capacity of the damper is at-
tained; yet if the damper does not enter Phase IV, the ultimate energy dissipation curve to
Metals 2021, 11, 183 26 of 30
be used is the solid red line of Figure 26; and if the damper enters Phase IV, the failure
curve is the dash red line.
6. Conclusions
This paper investigated a new metallic damper intended for use in protecting struc-
tures subjected to earthquakes. The damper has a gap mechanism that prevents high cycle
fatigue damage under wind loads. Seven identical specimens representing the damper
were tested under quasi-static and dynamic loadings on a shake table until failure. The
force-displacement curves were decomposed into the skeleton and Bauschinger parts to
verify the shape of the hysteretic curves and the ultimate energy dissipation capacity of
the damper. The following conclusions are reached:
1. The damper presents a very stable hysteretic response until failure, without any
sign of strength or stiffness degradation.
2. The damper features a multi-phased behavior. In Phase I the damper is not acti-
vated because the gap is not closed. In Phase II the damper remains in the elastic range.
In Phase III the damper dissipates energy through plastic deformations, keeping the re-
storing force approximately constant. In Phase IV the damper exhibits a significant in-
crease in strength and stiffness, and keeps dissipating energy. The damper can be de-
signed to remain in Phase I for wind loads, in Phase II for frequent earthquakes, in Phase
III for the “design earthquake”, and in Phase IV for the “maximum credible earthquake”.
3. The ultimate energy dissipation capacity of the damper if it enters Phase IV is about
35% larger that if it remains in Phase III.
4. A simple polygonal hysteretic model is proposed to predict reasonably well the
response of the damper under arbitrarily applied cyclic loads; it is able to capture the sig-
nificant increase in strength and stiffness in the large deformation range.
5. Equations defining the ultimate energy dissipation capacity of the dampers are
proposed, together with a criterion which, if applied in conjunction with the proposed
hysteretic model, allows one to reliably and accurately predict the failure of the damper.
6. A reinforced concrete structure equipped with the new metallic dampers was
tested on a shake table under realistic dynamic seismic loadings. During the preliminary
low intensity white noise signals applied for training the shake table, the dampers did not
become activated (i.e., they remained within Phase I). In the seismic tests representing a
frequent earthquake, the damper remained elastic (i.e., within Phase II). Under the seismic
test representing the “design earthquake”, the dampers experienced severe plastic defor-
mations but did not exhibit any significant associated increase in strength/stiffness (i.e.,
they remained in Phase III), and they kept the main structure basically undamaged. Under
the seismic tests that represented “maximum credible earthquakes”, the dampers exhib-
ited a significant increase of strength and stiffness, entering Phase IV; this protected the
main structure, limiting the damage and preventing collapse.
Data Availability Statement: Data available on request due to restrictions e.g., privacy or ethical.
Appendix A
Figure A1 shows the approximated pentalineal skeleton curve in a given domain of
loading (positive or negative) expressed in nondimensional form with the coefficients de-
fined in Equations (3) and (6)–(9). Figure A2 shows with shaded areas a detail of the en-
ergy dissipated in the range 0 − 𝜂 (Figure A2a) and in the range 𝜂 − 𝜂 (Figure
A2b). For the range of deformation on the skeleton part 𝜂 − 𝜂 a Figure similar to
Figure A2b can be drawn replacing 𝜏 with 𝜏 and 𝑘 with 𝑘 . For the range of de-
formation on the skeleton part 𝜂 > 𝜂 , a Figure similar to Figure A2b can be drawn
replacing 𝜏 with 𝜏 and 𝑘 with 𝑘 .
Focusing on Figure A2a, the relation between a given ordinate x and 𝜂 is:
(𝑥 − 1)
𝜂= − (𝑥 − 1) (A1)
𝑘
and solving for x gives:
𝑘 𝜂 + (1 − 𝑘 )
𝑥= (A2)
(1 − 𝑘 )
The shaded area in Figure A2a can be expressed by:
( )
𝜂= (A3)
For 𝜂≤ 𝜂 : 𝜂= 𝜂+ 𝜂 (A4)
( )
( ) ( ) ( )
For 𝜂 ≤ 𝜂 ≤ 𝜂 : 𝜂= + 2𝜏 + (A8)
( )
For 𝜂 > 𝜂 :
𝜂 (1 + 𝜏 ) ( 𝜂 − 𝜂 ) 𝑘 ( 𝜂 − 𝜂 )
𝜂= + 2𝜏 +
2 2 (1 − 𝑘 )
( 𝜂 − 𝜂 ) 𝑘 ( 𝜂 − 𝜂 )
+ 2𝜏 + (A10)
2 (1 − 𝑘 )
( 𝜂 − 𝜂 ) 𝑘 ( 𝜂 − 𝜂 )
+ 2𝜏 +
2 (1 − 𝑘 )
Noting that under cyclic loading the damper dissipates energy in the positive and in
the negative domains of loading and assuming that 𝛿 = 𝛿 , the total energy dissi-
pated on the skeleton part is two times that expressed by Equations (A4), (A8)–(A10), and
this gives the Equations (6)–(9) in Section 4.2.
(a) (b)
Figure A2. Detail of the energy dissipated in the range: (a) 0 − 𝜂 and (b) 𝜂 − 𝜂 .
Metals 2021, 11, 183 29 of 30
References
1. Skinner, R.I.; Kelly, J.M.; Heine, A.J. Hysteretic dampers for earthquake-resistant structures. Earthq. Eng. Struct. Dyn. 1974, 3,
287–296.
2. Soong, T.T.; Spencer, B.F., Jr. Supplemental energy dissipation: State-of-the-art and state-of-the-practice. Eng. Struct. 2002, 24,
243–259.
3. Constantinou, M.; Soong, T.; Dargush, G. Passive Energy Dissipation Systems for Structural Design and Retrofit; Monograph No. 1;
MCEER: Buffalo, NY, USA, 1998.
4. Symans, M.D.; Constantinou, M.C. Semi-active control systems for seismic protection of structures: A state-of-the-art review.
Eng. Struct. 1999, 21, 469–487.
5. Housner, G.W.; Bergman, L.A.; Caughey, T.K.; Chassiakos, A.G. Structural control: Past, present, and future. J. Eng. Mech. 1997,
123, 897–971.
6. Chan, R.; Albermani, F. Experimental study of steel slit damper for passive energy dissipation. Eng. Struct. 2008, 30, 1058–1066.
7. Javanmardi, A.; Ibrahim, Z.; Ghaedi, K.; Ghadim, H.B.; Hanif, M.U. State-of-the-art review of metallic dampers: Testing, devel-
opment and implementation. Arch. Comp. Meth. Eng. 2020, 27, 455–478.
8. Whittaker, A.S.; Bertero, V.V.; Thompson, C.L.; Alonso, L.J. Seismic testing of steel plate energy dissipation devices. Earthq.
Spect. 1991, 7, 563–604.
9. Tsai, K.; Chen, H.; Hong, C.; Su, Y. Design of steel triangular plate energy absorbers for seismic-resistant construction. Earthq.
Spect. 1993, 9, 505–528.
10. Kobori, T.; Miura, Y.; Fukuzawa, E.; Yamada, T.; Arita, T.; Takenaka, Y.; Miyagawa, N.; Tanaka, N.; Fukumoto, T. Development
and application of hysteresis steel dampers. In Proceedings of Tenth World Conference on Earthquake Engineering, Rotterdam,
The Netherlands, 19–24 July 1992; pp. 2341–2346.
11. Benavent-Climent, A.; Oh, S.; Akiyama, H. Ultimate energy absorption capacity of slit-type steel plates subjected to shear de-
formations. J. Struct. Constr. Eng. 1998, 63, 139–147.
12. Watanabe, A.; Hitomi, Y.; Saeki, E.; Wada, A.; Fujimoto, M. Properties of brace encased in buckling-restraining concrete and
steel tube. In Proceedings of Ninth World Conference on Earthquake Engineering, Tokyo-Kyoto, Japan, 2–9 August 1988; pp.
719–724.
13. Oh, S.; Kim, Y.; Ryu, H. Seismic performance of steel structures with slit dampers. Eng. Struct. 2009, 31, 1997–2008.
14. Ghabraie, K., Chan, R.; Huang, X.; Xie, Y.M. Shape optimization of metallic yielding devices for passive mitigation of seismic
energy. Eng. Struct. 2010, 32, 2258–2267.
15. Zheng, J.; Li, A.; Guo, T. Analytical and experimental study on mild steel dampers with non-uniform vertical slits. Earthq. Eng.
Eng. Vibr. 2015, 14, 111–123.
16. Lee, C.; Ju, Y.K.; Min, J.; Lho, S.; Kim, S. Non-uniform steel strip dampers subjected to cyclic loadings. Eng. Struct. 2015, 99, 192–
204.
17. Amiri, H.A.; Najafabadi, E.P.; Estekanchi, H.E. Experimental and analytical study of block slit damper. J. Constr. Steel Res. 2018,
141, 167–178.
18. Shao, F.; Gu, T.; Jia, L.; Ge, H.; Taguchi, M. Experimental study on damage detectable brace-type shear fuses. Eng. Struct. 2020,
225, 111260.
19. Benavent-Climent, A. A brace-type seismic damper based on yielding the walls of hollow structural sections. Eng. Struct. 2010,
32, 1113–1122.
20. Lee, J.; Kim, J. Development of box-shaped steel slit dampers for seismic retrofit of building structures. Eng. Struct. 2017, 150,
934–946.
21. Marshall, J.D.; Charney, F.A. Seismic response of steel frame structures with hybrid passive control systems. Earthq. Eng. Struct.
Dyn. 2012, 41, 715–733.
22. Lee, C.; Kim, J.; Kim, D.; Ryu, J.; Ju, Y.K. Numerical and experimental analysis of combined behavior of shear-type friction
damper and non-uniform strip damper for multi-level seismic protection. Eng. Struct. 2016, 114, 75–92.
23. Hashizume, S.; Takewaki, I. Hysteretic-viscous hybrid damper system with stopper mechanism for tall buildings under earth-
quake ground motions of extremely large amplitude. Front. Built Environ. 2020, 6, 1–16.
24. Fang, Z.; Li, A.; Li, W.; Shen, S. Wind-induced fatigue analysis of high-rise steel structures using equivalent structural stress
method. Appl. Sci. 2017, 7, 71.
25. Repetto, M.P.; Solari, G. Wind-induced fatigue collapse of real slender structures. Eng. Struct. 2010, 32, 3888–3898.
26. West, M.A.; Fisher, J.M.; Griffis, L.G. Serviceability Design Considerations for Steel Buildings; Design Guide 3; AISC: Chicago, IL,
USA, 2003.
27. Krawinkler, H. Guidelines for Cyclic Seismic Testing of Components of Steel Structures; ATC: Redwood City, CA, USA, 1992.
28. Benavent-Climent, A.; Donaire-Avila, J.; Oliver-Saiz, E. Seismic performance and damage evaluation of a waffle-flat plate struc-
ture with hysteretic dampers through shake-table tests. Earthq. Eng. Struct. Dyn. 2018, 47, 1250–1269.
29. Structural Engineers Association of California (SEAOC) & Vision 2000 Committee. Vision 2000: Performance Based Seismic Engi-
neering of Buildings; SEAOC & California Office of Emergency Services: Sacramento, CA, USA, 1995; Volume 1.
Metals 2021, 11, 183 30 of 30
30. Benavent-Climent, A. An energy-based damage model for seismic response of steel structures. Earthq. Eng. Struct. Dyn. 2007,
36, 1049–1064.
31. Kato, B.; Akiyama, H.; Yamanouchi, H. Predictable properties of structural steels subjected to incremental cyclic loading. In
IABSE Symposium on Resistance and Ultimate Deformability of Structures Acted on by Well Defined Loads; International Association
for Bridge and Structural Engineering: Lisbon, Portugal, 1973; pp. 119–124.